Contents
How do I run global setup/teardown only if tests will be run?
How do I clean up global state between running different tests?
Why cannot I derive from the built-in reporters?
What is Catch2's ABI stability policy?
What is Catch2's API stability policy?
Does Catch2 support running tests in parallel?
Can I compile Catch2 into a dynamic library?
What repeatability guarantees does Catch2 provide?
My build cannot find catch2/catch_user_config.hpp, how can I fix it?
Write a custom event listener and place the
global setup/teardown code into the testRun* events.
Write a custom event listener and place the
cleanup code into either testCase* or testCasePartial* events,
depending on how often the cleanup needs to happen.
They are not made to be overridden, in that we do not attempt to maintain a consistent internal state if a member function is overridden, and by forbidding users from using them as a base class, we can refactor them as needed later.
Catch2 provides no ABI stability guarantees whatsoever. Catch2 provides rich C++ interface, and trying to freeze its ABI would take a lot of pointless work.
Catch2 is not designed to be distributed as dynamic library, and you should really be able to compile everything with the same compiler binary.
Catch2 follows semver to the best of our ability. This means that we will not knowingly make backwards-incompatible changes without incrementing the major version number.
Not natively, no. We see running tests in parallel as the job of an external test runner, that can also run them in separate processes, support test execution timeouts and so on.
However, Catch2 provides some tools that make the job of external test runners easier. See the relevant section in our page on best practices.
Yes, Catch2 supports the standard CMake BUILD_SHARED_LIBS
option.
However, the dynamic library support is provided as-is. Catch2 does not
provide API export annotations, and so you can only use it as a dynamic
library on platforms that default to public visibility, or with tooling
support to force export Catch2's API.
There are two places where it is meaningful to talk about Catch2's repeatability guarantees without taking into account user-provided code. First one is in the test case shuffling, and the second one is the output from random generators.
Test case shuffling is repeatable across different platforms since v2.12.0, and it is also generally repeatable across versions, but we might break it from time to time. E.g. we broke repeatability with previous versions in v2.13.4 so that test cases with similar names are shuffled better.
Since Catch2 3.5.0 the random generators use custom distributions, that should be repeatable across different platforms, with few caveats. For details see the section on random generators in the Generator documentation.
Before this version, random generators relied on distributions from
platform's stdlib. We thus can provide no extra guarantee on top of the
ones given by your platform. Important: <random>'s distributions
are not specified to be repeatable across different platforms.
catch2/catch_user_config.hpp is a generated header that contains user
compile time configuration. It is generated by CMake/Meson/Bazel during
build. If you are not using either of these, your three options are to
- Build Catch2 separately using build tool that will generate the header
- Use the amalgamated files to build Catch2
- Use CMake to configure a build. This will generate the header and you can copy it into your own checkout of Catch2.